Hacker News new | past | comments | ask | show | jobs | submit login

Granted you can get a performance boost from denormalizing, and even maintain integrity, I still find the denormalization I want today won't be the denormalization I want tomorrow.

Moving from denormalization to denormalization can be hard, and you end up with a sui generis application, versus a normalized database where so much has been written about how to deal with problems up to a reasonably large scale, although the document store style does seem to be reasonably popular so maybe I'm missing something.

Depending on what I've needed in the past, I've simply duplicated all data from the RDBMS into denormalized structures for ElasticSearch or MongoDB. It's worked incredibly well for scenarios that are mostly-read where normalized queries are very expensive.

Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact